Microbial and chemical transformations of euphorbia factor L1 and the P-glycoprotein inhibitory activity in zebrafishes

Abstract Euphorbia factor L1 (EFL1, 1) is a natural tri-ester of 6,17-epoxylathyrol with cancer multidrug resistance (MDR) reversal activity. Several EFL1 derivatives (2–9) were prepared by chemical and microbial transformations and their ability to inhibit P-glycoprotein (P-gp) activity was estimated. Six de-acylated derivatives (2–7) were obtained through base-hydrolysis of 1, and the base-catalysed hydrolysis via KOH and NaOH yielded different hydrolysed products of 1. Two biotransformed products (8–9) were directly obtained via microbial transformation of 1, and 8 was also formed by microbial conversion of the hydrolysed product 3. The P-gp modulation of 1–9 was assessed by a zebrafish model. The substrate 1 and its biotransformed product 9 as the tri-esters of lathyranes possessed the highest P-gp inhibitory activity with IC50 values of 34.97 and 15.50 µM, respectively, through down-regulating P-gp expression at the protein level rather than at MDR1 mRNA level. Graphical Abstract


Introduction
Multidrug resistance (MDR) is a leading cause of failure in cancer chemotherapy, and one of the major mediators is the multidrug extrusion pump protein, P-glycoprotein (P-gp) (Sharom 1997). Therefore, P-gp inhibitors can be considered as promising lead agents to overcome MDR (Zhang et al. 2021). Various natural and synthesised lathyrane polyesters are frequently reported with MDR reversal activity Reis et al. 2020;Yang et al. 2020) through P-gp inhibition (Duarte et al. 2007; Barile et al. 2008). The MDR reversal activity of lathyranes is known to be greatly influenced by acylation of hydroxyl groups positioned on diterpenol skeletons (Jiao et al. 2015). In the past decades, various attempts have been made by acylation of lathyranes with different acylation sites as well as types and numbers of acyl groups (Vieira et al. 2014;Jiao et al. 2015;Matos et al. 2015). However, the role of the lathyrane skeleton itself in improving MDR reversal activity is less clear. In addition, in a recent study conducted in our laboratory, various new lathyrane derivatives (such as cyclopropaneopening rearrangement, and hydroxylation at non-activated carbons) formed by microbial transformation greatly improved lathyrane structural diversity and MDR reversal activity (Wu et al. 2016). These derivatives will facilitate a better understanding of the lathyrane skeleton instead of acylated side-chains on MDR reversal activity.
Euphorbia factor L1 (EFL1, 1) is a natural tri-ester of 6,17-epoxylathyrol in Euphorbia lathyris L., and possesses MDR reversal activity through P-gp inhibition (Zhang et al. 2013). Previous studies have focused on structural modification of Euphorbia factor L3 and 5,6-epoxy lathyranes (Jiao et al. 2015;Matos et al. 2015;Wu et al. 2016;Wang et al. 2020). Very few works concerning the improving structural diversities of 6,17-epoxy lathyranes have been published so far (Matos et al. 2015;Monico et al. 2017). In this study, enhancing structural diversity of lathyranes by chemical and microbial transformations will be illustrated using an example of 1. Nine lathyrane derivatives including six hydrolysed products (2-7) and two biotransformed derivatives (8 and 9) were prepared and identified. The MDR reversal activity of products was evaluated in a zebrafish model through targeting P-gp and MDR1 gene.

Results and discussion
2.1. Chemical transformation of EFL1 (1) through base hydrolysis Originally, we attempted to obtain the total de-esterified product (6,17-epoxylathyrol) of 1 by NaOH-catalysed hydrolysis. However, the hydrolysis occurred in a timedependent manner, and 6,17-epoxylathyrol was not the only hydrolysed product as expected. In a further attempt to maximise product yields and to understand the effect of base nature on hydrolysed products, the hydrolysis reaction was also performed in KOH. The time-dependent changes of relative amounts of four hydrolysed products of EFL1 (1) in NaOH-and KOH methanol solutions were obtained by HPLC-DAD ( Figure S1). These two hydrolysis reactions resulted in different product profiles and formation of only one common product (peak 3). These products were then prepared and identified by NMR and MS spectra (Figure 1). Products 2 À 4 and 3/5 À 7 were obtained from the KOH and NaOH hydrolysates of 1, respectively. Among these, products 2, 4, 5, and 7 were new compounds.
Product 2 (5 mg, 5% yield) was assigned as a molecular comparison of NMR data between 2 and 1 (Tables S1 and S2) showed a complete loss of C-3, C-5 and C-15 acyl groups in 2, accompanied by the appearance of a new oxygenated methyl carbon at d C 59.2, and a characteristic downfield shift of C-17 (d C 66.8) in the C-6,17 epoxy unit. This indicated 2 might be a mono-O-methylated C-6,17epoxy opening analoge (Matos et al. 2015). This assumption was supported by HMBC correlations from H-5 (d H 3.94) to C-4 (d C 54.3), C-6 (d C 78.2), and the newly appeared methoxy signal (d C 59.2) ( Figure S2), which pointed out that the cleavage occurred at the C-6,17-epoxy and the methyl located at O-5. Therefore, the structure of 2 was identified as 5-methoxy-6,17-seco-epoxylathyrol. Products 3 (30 mg, 30 and 40% yields in KOH-and NaOH hydrolysis, respectively; ESI-MS m/z 351 [M þ H] þ ) and 6 (40 mg, 40% yield; ESI-MS m/z 351 [M þ H] þ ) were identified as 6,17-epoxylathyrol and 7b-hydroxylathyrol, respectively, through comparison of their NMR and MS data with those published in the literature (Vieira et al. 2014;Wang et al. 2014 Table S2). The 13 C NMR spectrum of 4 also showed the complete disappearance of three acyl signals and a new oxygenised CH 3 carbon at d C 59.6, similarly as observed in 2. The most notable difference between the 13 C NMR spectra of 4 and 2 concerned signals attributed to the C-17 chemical shit: d C 77.9 in 4 was deduced to be 17-Omethylated substitution. This linkage was confirmed by HMBC correlations from H-17 (d H 3.41) to C-6 (d C 75.0) and the methoxy signal (d C 59.6) ( Figure S2). The structure of 4 was thus identified as 17-methoxy-6,17-seco-epoxylathyrol. This indicated that 4 underwent the same chemical transformation pathway as 2, a similar 6,17-epoxy ringopened transformation. Product 5 (4 mg, 4% yield) possessed a molecular formula of C 21 H 34 O 6 as determined by its HR-ESI-MS at m/z 383.2433 [M þ H] þ (calcd for C 21 H 35 O 6 þ , 383.2434) and 13 C NMR data (Table S2). In its 13 C NMR spectrum, the lack of ester carbonyl signals around 170.0 ppm indicated the complete disappearance of all three acyl groups. The NMR data of 5 were quite different from those of 1, indicating that 5 might possess a different structural skeleton. A detailed comparison between NMR data of 5 and ethoxyboetirol (a base rearrangement product of epoxyboetirane A) (Matos et al. 2015) suggested that their carbon signals were almost identical, except that a carbon signal corresponding to an oxygenised methyl (d C 60.0) in 5 was observed instead of an oxygenised ethyl group in ethoxyboetriol. The HMBC correlations from H-5 (d H 3.95) to C-3 (d C 74.7), C-4 (d C 54.5), C-15 (d C 210.6), C-17 (d C 66.2), and the newly occurring carbon signal (d C 60.0) confirmed the occurrence of methyl at OH-5 ( Figure  S2). The structure of 5 was thus identified as methoxyboetirol.
Product 7 (5 mg; 5% yield) had a molecular formula of C 20 H 32 O 6 deduced by HR- þ , 369.2199) and its 13 C NMR data (Table S2). The NMR spectra of 7 also showed signals for a tri-de-esterified derivative of 1. A comparison between the 13 C NMR data of 7 and 3 revealed that their carbon signals were almost identical, except for the downfield shifting C-17 and neighbor carbons. This suggested a C-6,17 epoxy ring-opened structure, supported by the HMBC spectrum ( Figure S2). The structure of 7 was thus identified as 6,17-seco-epoxylathyrol.
Based on the identification of hydrolysed products and the time-dependent reaction profiles, several conclusions might be drawn from the base hydrolysis of 1: 1) Both the base-catalysed reactions (24 h) formed four hydrolysed products, and only 3 was found as their common product. 2) The hydrolysis either in KOH or NaOH was finished within 2 h, as evident from the disappearance of 1. 3) No mono-ester or di-ester of lathyrol derivatives were produced from 1, indicating that the base hydrolysis was not regio-selective. 4) The hydrolysis in KOH resulted in production of a higher ratio of 3 regardless of reaction time, though other de-esterfied products (such as 2 and 4) increased slightly with increasing reaction times. However, the NaOH-catalysed hydrolysis afforded two main products (3 and 6) that might be inter-converted, as evident from the time-dependent change of their relative contents. And 5) The product 5 and its derivative, as the structural rearrangement product of 6,17-epoxylanthyrols can be obtained in both KOH- (Matos et al. 2015) and NaOH-catalysed hydrolysis.
Several natural lathyrane polyesters were reported to be hydrolysed, and the obtained de-acylated lathyranes were re-acylated back with a variety of acylated groups (Reis et al. 2013;Jiao et al. 2015;Matos et al. 2015;Monico et al. 2017;Reis et al. 2017). Thus, achievement of native diterpenols of polyesters is important for subsequent structural reconstruction of synthetic polyesters. Our study demonstrated that KOH was suitable for hydrolysis of 1 with a relative higher yield (only 9% in literature) (Appendino et al. 2000), whereas NaOH-catalysed hydrolysis facilitated formation of another main lathyrol, 7b-hydroxylathyrol (6). The other unexpected diterpenols (2, 4, 5 and 7) will afford further synthesis opportunity using them as starting materials, based on which the effect of different diterpenol structures on MDR reversal activity can be explored.

Microbial transformation of EFL1 (1) and 6,17-epoxylathyrol (3)
A total of 28 bacteria were screened for their ability to utilise 1 and 3. Two species (Mucor polymorphosporus and Cunninghamella elegans) were found to biotransform 1, and only M. polymorphosporus was found to transform 3 in our experimental conditions. Preparative-scale biotransformation of 1 afforded two metabolites (8 and 9), and bioconversion of 3 led to production of 8 as its only metabolite (Figure 1). Structures of the biotransformed products were identified via NMR and MS spectra.
Compound 8 (4 mg, 5% yield) was isolated from the culture of M. polymorphosporus co-incubated with the substrate 1, and identified as lathyrol (Jiao et al. 2010). This metabolite (8, 3 mg, 4% yield) was also obtained by M. polymorphosporus co-incubated with 3. Compound 9 (6 mg, 4% yield) was obtained from the culture of C. elegans incubated with 1 (Figure 1), and identified as deoxy Euphorbia factor L1 (deoxy EFL1) (Appendino et al. 1999). The substrates EFL1 (1) and 3 underwent the reduction of the epoxy ring to exocyclic double bond by M. polymorphosporus, indicating that the bacterium might have some enzymes responsible for this conversion. Metabolite 8 also experienced a further tri-de-esterification. Microbial transformation is known to be highly substrate-specific, since hydroxylation, rearrangement, and acetylation of some lathyrane diterpenoids (such as lathyrol and EFL3) were observed by some of these bacteria (Wu et al. 2016). This is the first report on structural modification of EFL1 and its de-acylated form (6,17-epoxylathyrol) by microbial transformation.

Biological activity on P-gp inhibition by transformed products of EFL1 (1)
The zebrafish embryo (Danio rerio) is an important model organism and used in various drug evaluation. It is reported that all ATP-binding cassette protein subfamilies including P-gp in zebrafishes are found to coincide with the human subfamilies (Annilo et al. 2006). P-gp, the product of MDR1 gene, is a transmembrane efflux pump for various anticancer drugs. Thus, zebrafish is a suitable experimental model for assessing P-gp inhibitory activity. In this model, zebrafishes are exposed to rhodamine B (a fluorescent substrate for P-gp), which produces a weak fluorescence response inside zebrafishes due to the presence of P-gp. When co-treated with P-gp inhibitors, zebrafishes undergo a significant fluorescence enhancement (Ko et al. 2011). This fluorescent intensity can be quantitatively recorded by a microscope, and the data are expressed as percentage relative to the vehicle control group.
Due to very limited amounts of the transformed compounds, only 1 was selected for maximum tolerated concentration (MTC) determination for the lathyrane derivatives. As a result, zebrafishes treated with 10.4-50.0 mM of 1 were normal but tiny precipitation displayed at concentrations of 12.5-50.0 lM. Thus, it was reasoned that 50.0 lM was the MTC with 100% of zebrafish survival.

P-Gp inhibitory effect of 1-9 in zebrafishes
Compounds 1-9 were estimated for their P-gp inhibitory activity in the zebrafish model. As shown in Figures S3 and S4, the fluorescence imaging of cyclosporin A exhibited higher fluorescent intensity (1825531 in pixel) compared to that of vehicle control (1382983 in pixel, p < 0.001). This indicated a 37% inhibition of P-gp by cyclosporin A, and that the zebrafish model was suitable for evaluation of P-gp inhibition. Subsequently, compounds 1-9 at a concentration of 50 mM were assessed for the P-gp inhibitory effect. The results reveled that groups treated with 1, 3, 7-9 (p < 0.001), 5 (p < 0.01), and 6 (p < 0.05) exhibited significant P-gp inhibition with inhibitory percentage of 21-116% compared to vehicle control ( Figure S4). However, 2 and 4 with a methoxy substitution showed no obvious P-gp inhibitory effect (p > 0.05). These data confirmed that lathyrane derivatives with acyl groups (1 and 9) would greatly enhance the P-gp inhibitory activity. In addition, seven deacylated products (2-8) displayed totally different activity ( Figure S4), indicating that fine structures of lathyrane skeletons might affect the P-gp inhibitory activity. The transformed products (1 and 9) with higher P-gp inhibitory activity (> 50% at 50 lM) were selected for further IC 50 determination ( Figure S5). The IC 50 values for 1 and 9 were 34.97 and 15.50 mM, respectively. Our study demonstrated that the biotransformed product 9 with a 6,17-exocyclic double bond possessed a higher P-gp inhibition than 1 with a 6,17-epoxy group, which was highly consistent with the finding of Jiao et al. (2009). 2.3.2. Inhibitory effect of 1 and 9 on MDR1 gene and P-gp expression in zebrafishes The MDR1, the gene coding P-gp, may regulate the expression and activity of P-gp. Thus, EFL1 (1) and its biotransformed product 9 were selected to study their effect on MDR1 gene expression by RT-PCR analysis. As shown in Figure S6A, the increase of MDR1 mRNA by cyclosporin A was 3.4-fold compared to vehicle control, indicating that cyclosporin A has significant effect on induction of MDR1 mRNA in zebrafishes. EFL1 acted similarly by up-regulating MDR gene expression. However, the MRD1 mRNA levels following treatment with 9 resulted in significant decrease compared to those of cyclosporin A and 1, indicating that the MDR1 gene expression was potentially repressed by 9. The result suggested that the MDR reversal activity of 9 increased compared to that of 1 through down-regulation of MDR1 gene expression.
Furthermore, Western blot was applied to assess the regulation effect of 1 and 9 on P-gp expression in zebrafishes. As shown in Figure S6B, exposure of zebrafishes to 1 and 9 (each 25 mM) for 12 h resulted in a significant decrease in P-gp level. P-gp expression levels in zebrafishes decreased by 71% and 72% after treatments with 25 mM of 1 and 9, respectively, as compared to vehicle control. Compared to deacylated products 2-8 (all with log p values less than 2.0), compounds 1 and 9 have higher log p values (3.5 and 4.5), which is consistent with the basic structural requirements for a better P-gp inhibitor (Jabeen et al. 2012). The above Western blot and RT-PCR analysis showed that 1 and 9 gave almost the same level of P-gp inhibition, and 9 had a higher inhibitory activity on MDR1 gene. This indicated that the biotransformed product 9 might have more potential against cancer MDR.

Base hydrolysis of EFL1 (1) and preparation of hydrolysed products (2-7)
EFL1 (1) was hydrolysed in either 5% KOH or NaOH methanol solution. Each hydrolysis was stirred at room temperature with reaction time of 2, 4, 6, 12, or 24 h. After reaction, the mixture was poured into an equal volume of pure water, and was extracted three times with EtOAc. The organic layer was collected and dried over anhydrous Na 2 SO 4 . Solvents were removed under vacuum, and the obtained residue was re-dissolved in methanol for HPLC analysis. For preparation of hydrolysed products, 1 (100 mg) was suspended and stirred into a 5% KOH or NaOH solution (10 mL) for 24 h. The EtOAc extract of the reaction mixture was purified by semi-preparative HPLC (Agilent 1260; C 18 column, 250 Â 10 mm, 5 lm; 40-80% linear gradient of MeOH/H 2 O in 40 min; flow rate, 3.0 mL/min; column temperature, 25 C; DAD, 280 nm).
Preparative-scale bio-transformations of 1 and 3 were performed in 1-L Erlenmeyer flasks containing 250 mL of stage-II culture medium, using the same incubation conditions as the small-scale screening experiments (Wu et al. 2016). Addition of 25 mL substrate solution (4 mg/mL in ethanol) allowed the biotransformation to proceed for 5 days. After the transformation was finished, the culture mixture was filtered. Biotransformed products were extracted from both the filtrate and the mycelia using EtOAc as described above. The combined EtOAc extracts of the mycelia and filtrate were purified as described above.

Determination of MTC in zebrafishes
Healthy fertilised zebrafishes (wild-type AB; 6 h post fertilisation) were obtained from Hangzhou Hunter Biotech Co., Ltd. (Hangzhou, China) with the license number SYXK 2012-0171. Zebrafishes were maintained at 28 C in reconstituted water (200 mg/L instant seat salts in reverse osmosis water; conductivity: 480-510 mS/cm; hardness: 53.7-71.6 mg/L as CaCO 3 ; pH: 6.9-7.2), and selected by an Olympus SZX7 stereomicroscope (Tokyo, Japan) for bio-estimation. MTC of samples were determined using 1 as an example. The MTC was defined as the maximum concentration at which no significant toxicity to zebrafishes. Thirty zebrafishes were randomly cultured in 6-well plates. Various concentrations of 1 (0.391-50.0 mM) were added to 6-well plates, followed by incubation for 18 h at 28 C. The amounts of cell death were measured under the stereomicroscope. Zebrafishes treated with reconstituted water were used as vehicle control.

P-Gp inhibitory effect of EFL1 derivatives in zebrafishes
Stocking solutions of samples (50 mM) and cyclosporin A (20 mM) were prepared in DMSO ( 0.1% final concentration), and a series of diluted solutions were prepared with reconstructed water. Healthy zebrafishes (30 eggs/well) were randomly divided into 6-well plates and treated with 50 lM of 1-9. After incubation at 28 C for 18 h, Rhodamine B (5 lM) was added to each well, and 10 zebrafishes from each well were randomly selected for observation and measurement of fluorescent intensity (FI). Zebrafish images were recorded by a Nikon AZ100 multizoom microscope, and the quantitative image analysis was conducted by measuring FI with Nikon NIS-Elements D.310 advanced image processing software (Tokyo, Japan). Cyclosporin A and reconstituted water were used as positive and vehicle controls, respectively. The level of P-gp inhibition was calculated as: 3.4.3. Effect of EFL1 (1) and 9 on MDR1 gene expression in zebrafishes Zebrafishes were placed (30 eggs/well) in 6-well plates, and treated with 25 lM of 1 and 9 for 24 h. Zebrafishes treated with reconstituted water and cyclosporin A (20 mM) served as vehicle and positive controls, respectively. Total RNAs of treated zebrafishes were extracted using the TRIzol method. The purity and quality of total RNAs were estimated by OD measurements using a Thermo Multiskan MK3 microplate reader (Shanghai, China). The ratio of OD at 260 and 280 nm (A260/A280) was taken as a measure of purity, with pure RNA having an A260/A280 of 1.81 À 1.96. The first-strand cDNA was synthesised using 2 lg total RNAs with a FastQuant RT kit (with gDNase) according to the manufacturer's instruction. The cDNA (20 mL) mixture was kept at À20 C until further qPCR analysis with Bio-Rad CFX Connect Real-Time PCR detection system. Primers for MDR1: MDR1-F (5 0 -CCT GGC TTC TCA ATC TCA T) and MDR1-R (5 0 -TTT ACT ACT CCC TTG TAA CGC). Primers for b-actin: b-actin-F (5 0 -TCG AGC AGG AGA TGG GAA CC) and b-actin-R (5 0 -CTC GTG GAT ACC GCA AGA TTC). The relative quantification of MDR1 gene expression was conducted using the 2 -᭝᭝Ct method and normalised to the expression of b-actin mRNA.

Effect of EFL1
(1) and 9 on P-gp expression in zebrafishes Zebrafishes in 6-well plates (30 eggs/well) were treated with 25 lM of 1 and 9 for 24 h, lysed with ice-cold RIPA lysis buffer, and centrifuged at 12,000 g for 5 min. The supernatant was collected and subjected to Western blot with b-actin as a loading control. Zebrafishes treated with reconstituted water and cyclosporin A (20 mM) were used as vehicle and positive controls, respectively. P-gp expression was expressed as the intensity ratio of P-gp band to b-actin band in the same blot.

Statistical analysis
Biological experiments were performed in triplicate. The data were expressed as means ± SD. Significance of differences between groups was determined by one-way analysis of variance (ANOVA) followed by a Dunnett's test. A p-value of < 0.05 was considered statistically significant.

Conclusion
In this study, six hydrolysed and two biotransformed products including four new ones were prepared from EFL1. The P-gp inhibitory activity of the transformed compounds were assessed in a zebrafish mode. The biotransformed compound, deoxy EFL1 possessed the highest P-gp inhibitory activity through down-regulating P-gp expression at the protein level.

Disclosure statement
No potential conflict of interest was reported by the authors.

Funding
This work was supported by the grant from the Natural Science Foundation of Shanghai (No. 19ZR1406100).